neuropsychologia - university of victoriaweb.uvic.ca/.../exhaustingexercise.pdf · 2015-01-07 ·...

Pushing to the limits: The dynamics of cognitive control duringexhausting exercise

Cyril Schmit a,b,n, Karen Davranche d, Christopher S. Easthope a,c, Serge S. Colson a,Jeanick Brisswalter a, Rémi Radel aQ1

a Laboratoire LAMHESS (EA6309), Université de Nice Sophia-Antipolis, Franceb Institut National du Sport, de l’Expertise et de la Performance (INSEP), Département de la Recherche, Paris, Francec Spinal Cord Injury Centre, Balgrist University Hospital, Zurich, Switzerlandd Aix Marseille Université, CNRS, LPC UMR 7290, FR 3C FR 3512, 13331 Marseille cedex 3, France

a r t i c l e i n f o

Article history:Received 26 May 2014Received in revised form25 November 2014Accepted 7 January 2015

Keywords:Cognitive functionsResponse inhibitionPhysical exerciseCerebral oxygenationExhaustion

a b s t r a c t

This study aimed at investigating concurrent changes in cognitive control and cerebral oxygenation (Cox)during steady intense exercise to volitional exhaustion. Fifteen participants were monitored using pre-frontal near-infrared spectroscopy and electromyography of the thumb muscles during the completion ofan Eriksen flanker task completed either at rest (control condition) or while cycling at a strenuous in-tensity until exhaustion (exercise condition). Two time windows were matched between the conditionsto distinguish a potential exercise-induced evolutive cognitive effect: an initial period and a terminalperiod. In the initial period, Cox remained unaltered and, contrary to theoretical predictions, exercise didnot induce any deficit in selective response inhibition. Rather, the drop-off of the delta curve as reactiontime lengthened suggested enhanced efficiency of cognitive processes in the first part of the exercisebout. Shortly before exhaustion, Cox values were severely reduced – though not characteristic of a hy-pofrontality state – while no sign of deficit in selective response inhibition was observed. Despite this,individual’s susceptibility to making fast impulsive errors increased and less efficient online correction ofincorrect activation was observed near exhaustion. A negative correlation between Cox values and errorrate was observed and is discussed in terms of cerebral resources redistribution.

& 2015 Published by Elsevier Ltd.

1. Introduction

It is now well-accepted that physical exercise has a positiveeffect on basic cognitive functions (Tomporowski, 2003; Lam-bourne and Tomporowski, 2010), however its impact on highercognitive processes (e.g. selective inhibition) is less clear. Accord-ing to the different meta-analyzes on exercise and cognition, itseems that exercise would lead to a rather small positive effect oncognitive functioning due to large variations of the reported re-sults (Chang et al., 2012; McMorris et al., 2011). It has been pro-posed that exercise intensity would be the most important factorto explain this variability. Specifically, an inverted-U function hasbeen suggested in such a form that exercise above a certain in-tensity is no longer beneficial to cognitive functioning (McMorrisand Hale, 2012). In other words, while moderate exercise is asso-ciated with a positive effect, intense exercise (i.e. above the second

ventilatory threshold, VT2) is associated to a null or negative effecton cognitive functioning. This view has been supported by themain theoretical models. According to the arousal-cognitive per-formance (Yerkes and Dodson, 1908), the catecholamine (Cooper,1973; McMorris et al., 2008) and the reticular-activating hypo-frontality (Dietrich, 2003, 2009; Dietrich and Audiffren, 2011)theories, intense exercise, by inducing high levels of arousal, in-creasing neural noise, or down-regulating prefrontal cortex activ-ity, respectively, is predicted to impede higher cognitive processes.

Evidence of the detrimental effect of intense exercise mostlycome from studies using incremental protocols in which the ef-fects of exercise, probed at the end of the exercise, are confoundedwith the effect of exhaustion (e.g. Ando et al., 2005; Chmura andNazar, 2010; McMorris et al., 2009). The present study aimed todissociate the effect of the intensity from the state of exhaustionusing a steady state intense exercise performed until exhaustion.Exhaustion is a psychophysiological state concluding a fatiguedevelopment process. It is a predictable consequence to anystrenuous exercise. The state of exhaustion can also be consideredas the time spent on the task. In a concomitant realization of achoice reaction time task and a fatiguing submaximal contraction,

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

6768697071727374757677787980818283848586878889909192

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/neuropsychologia

Neuropsychologia

http://dx.doi.org/10.1016/j.neuropsychologia.2015.01.0060028-3932/& 2015 Published by Elsevier Ltd.

n Correspondence to: Institut National du Sport, de l’Expertise et de la Perfor-mance. 11, avenue du Tremblay, 75 012 Paris, France.

E-mail address: [email protected] (C. Schmit).

Please cite this article as: Schmit, C., et al., Pushing to the limits: The dynamics of cognitive control during exhausting exercise.Neuropsychologia (2015), http://dx.doi.org/10.1016/j.neuropsychologia.2015.01.006i

Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Lorist et al. (2002) showed that the more the time-on-taskselapsed, the more error rate and force variability of hand musclesincreased, suggesting an increasing detrimental effect of the dual-task on the dual-performance.

The detrimental effect of exercise on cognition occurring nearexhaustion can be explained as neural competition for the limitedcerebral resources between different centers of the brain (Dietrich,2003, 2009; Dietrich and Audiffren, 2011). Specifically, the hypo-frontality theory predicts that the prefrontal cortex (PFC), in re-sponse to the development of muscular fatigue, would be down-regulated to favor the allocation of resources to motor areas, whichwould in turn result in weaker cognitive functioning. The resourceredistribution hypothesis has been supported by animal studiesassessing cerebral activity during exercise. Tracing regional cere-bral blood flow and local cerebral glucose utilization as indexes ofcerebral activity showed a selective cortical and subcortical re-cruitment of brain areas during exercise (Delp et al., 2001; Grosset al., 1980; Holschneider et al., 2003; Vissing et al., 1996). Speci-fically, while motor and sensory cortices, basal ganglia, cerebellum,midbrain and brainstem nuclei were consistently activated at highintensities, the frontal cortex rather showed signs of deactivation.Positron emission tomography studies have also provided a similarpattern of findings in human (Tashiro et al., 2001, Kemppainenet al., 2005). If such neural balance may rationalize down-regu-lated cognitive performances during exercise (Dietrich and Audi-ffren, 2011), previous result remain based on an instantaneousdescription of the neural pattern during exercise, which does notinform about changes related to the proximity of exhaustion.

Near-infrared spectroscopy (NIRS) is a continuous tissue-monitoring technique which is able of tracking cerebral oxygena-tion (Cox) during exercise due to its relative robustness duringmovement (see Perrey, 2008). The NIRS method has been vali-dated and correlates highly with both electro-encephalographicand functional magnetic resonance imaging responses (Timinkulet al., 2010; Toronov et al., 2001). NIRS studies during exercisehave shown a very dynamic Cox pattern. A meta-analysis by Rookset al. (2010) proposes an intensity based account with the secondventilatory threshold as the critical reversing point in the in-verted-U relation between exercise intensity and PFC oxygenationdetermined through oxyhemoglobin concentration [HbO2].Nevertheless, this pattern is specific to untrained participants, astrained participants do not show any [HbO2] decline at high in-tensities. In addition, most of the studies reviewed employed in-cremental protocols. In studies that push participants until ex-haustion, results suggest that [HbO2] reduction may be related tothe individual time course to exhaustion. Timinkul et al. (2008)reported an individual timing of Cox desaturation, occurring be-fore VT2 for some participants and after for others. During steadyintense exhausting exercise, Shibuya et al. (2004a,b) reported Coxpatterns identical to those observed in incremental exercise (e.g.,Bhambhani et al., 2007; Rupp and Perrey, 2008). However, thepaucity of the exercise-NIRS studies on intense as well as pro-longed exercise to exhaustion in humans cannot ensure the va-lidity of a fatigue-based PFC deactivation hypothesis.

The second aim of the present study was to investigate con-current changes in cognitive performance and Cox in PFC whileperforming strenuous exercise until volitional exhaustion. Morespecifically, the protocol was designed to determine whethercognitive performance and Cox follow a similar dynamic. In aninitial period of intense exercise, it was anticipated that cognitiveperformance would be facilitated and Cox elevated compared tothe same initial period at rest. Then, in a critical period occurringjust before exhaustion, we expected a decrease in cognitive per-formance and a drop of PFC [HbO2] in comparison to the sameperiod at rest. Cognitive performances were assessed using amodified version of the Eriksen flanker task (Eriksen and Eriksen,

1974), consisting in overcoming the irrelevant dimension of thestimulus to give the correct response, to probe the efficiency ofselective response inhibition. The flanker task has been largelyused to investigate the effects of exercise on cognitive co Q2ntrol(Davranche et al., 2009a,b; McMorris et al., 2009; Pontifex andHillman, 2007). Although mean RT and average error rate doprovide valuable information relative to cognitive processes,more-detailed data analyzes uncover modulations that the soleconsideration of central tendency indices cannot reveal. Indeed,combined to RT distribution analyzes, conflict tasks have proved tobe powerful for assessing the processes implemented during de-cision-making tasks while exercising (Davranche and McMorris,2009; Davranche et al., 2009a,b; Joyce et al., 2014).

On a substantial amount of trials, although the correct responsewas given, a subthreshold electromyographic (EMG) activity in themuscles involved in the incorrect response could be observed.Such subthreshold EMG activities, named “partial errors”, reflectincorrect action impulses that were successfully corrected in orderto prevent a response error (Hasbroucq et al., 1999). To evaluatethe efficiency of the cognitive control during exercise, electro-myographic (EMG) activity of response effector muscles weremonitored to estimate the number of partial EMG errors. In orderto measure Cox, NIRS recording was centered on the right inferiorfrontal cortex (rIFC) as this brain area is a main region involved inthe brain network supporting the inhibition function (Aron et al.,2004, 2014), and has been described as the most responsive regionwhile performing the Eriksen flanker task ( Q3Hazeltine et al., 2000).Additionally, the distribution-analytical technique and the deltaplot analysis (Ridderinkhof, 2002; Ridderinkhof et al., 2004) wereused to assess the efficiency of cognitive control and the pro-pensity to make fast impulsive reactions through the analyzes ofthe percentage of correct responses (CAF) and the magnitude ofthe interference effect (delta curve) as a function of the latency ofthe response (van den Wildenberg et al., 2010). If exhausting ex-ercise impairs the efficiency of cognitive control, the drop-off ofthe delta curve should be less pronounced in the terminal periodthan in the initial period. If the propensity to commit impulsiveerrors increases before exhaustion, more errors are expected forfast RT trials on distributional analyzes of response errors.

2. Method

2.1. Participants

Fifteen volunteers took part in this experiment. They weremostly classified as untrained following the VO2̇ max criteria of dePauw et al. (2013) and had basic cycling experience (o1 h aweek). Informed written consent was obtained according to thedeclaration of Helsinki. Participants’ anthropometrical and

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132

Table 1Anthropometrical and physiological characteristics of participants.

Variables Mean 7 SDAll Women Men

Sample size 15 5 10Age [years] 22.170.6 23.271.2 21.570.6Height [cm] 175.972.6 166.371.8 180.772.6Body mass [kg] 66.573.1 52.972.1 73.372.3VO2̇ max [ml kg-1 min-1] 44.571.9 45.272.9 44.172.6

Maximal HR [bpm] 17872.6 18071.9 17773.0MAP [W] 261.3714.2 204.3710.3 290.2712.2

Results are presented as the mean group ± SD.Notes. SD¼standard deviation; MAP¼maximal aerobic power; VO2̇ max¼maximaloxygen consumption; HR¼heart rate.

C. Schmit et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎2


physiological characteristics are presented in Table 1.

2.2. Apparatus and display

All sessions were performed on a cycle ergometer (Brain-bikeNeuroActive, Motion Fitness, 1775 Winnetka Circle, Rolling Mea-dows, IL 60008) equipped with a handlebar and soft paddingsupports to comfortably support forearms. Two thumb responsekeys were fixed on the top of the right and left handle grips. Ascreen mounted on the ergometer at head height faced partici-pants at a mean distance of 70 cm.

2.3. Cognitive task

The cognitive task consisted of a modified version of the Erik-sen flanker task (Eriksen and Eriksen, 1974). Participants wererequired to complete 20 blocks of 40 trials during the controlsession and as many blocks as possible until exhaustion in theexercise session. There were two types of trials in each block:congruent trials (CO, 50%, all arrows uni-directional) and incon-gruent trials (IN, 50%, center arrow contra-directional). Each trialbegan with the presentation of a cross at the center as a fixationpoint. After 800 ms, the stimulus was presented and participantshad to respond according to the direction pointed by the centralarrow. The delivery of the response turned off the stimulus. Whenparticipants failed to respond within 2000 ms, the stimulus wasterminated and the next trial began. The inter-stimulus intervalwas 800 ms. In this modified version, each group of arrows couldrandomly be displayed either at the top or at the bottom of thescreen (Fig. 1). This modification ensured a higher processing ofthe flankers as participants could not anticipate the location of thecentral arrow.

2.4. Electromyographic measurement

Surface EMG activity of the flexor pollicis brevis (FPB) of eachthumb and of the vastus lateralis (VL) of the right thigh were re-corded using bipolar Ag/AgCl electrodes (diameter: 10 mm, inter-electrode distance (FPB, VL): 12 mm, 20 mm). The common re-ference electrode was situated on the head of the second meta-carpal of the right hand. Electrodes locations were determined asproposed by Hasbroucq et al. (1999) and were marked in order toensure congruent positioning in both sessions. Low inter-electrodeimpedance (o3 kΩ) was obtained via skin preparation. EMGsignals were amplified (gain x1000), filtered (pass-band 10–

500 Hz) and recorded at 2000 Hz (MP100 Biopacs Systems Inc.,Holliston, MA, USA). During both sessions, the experimenter con-tinuously monitored the signal of the thumbs and asked thesubjects to relax their muscles if the signal became noisy throughincreased preactivation.

2.5. Cerebral oxygenation measurement

Cox was monitored using a Near-infrared Spectroscopy (NIRS)system (Artinis Medical, PortaMon, The Netherlands). The NIRSevaluates the amount of infrared light that effectively traverses aninvestigated tissue from an emitter to a receptor. The receivedsignal allows the description of a quantitative change from base-line in chromophore concentration. The PortaMon continuouslyemits two wavelengths of IR light, 773 and 853 nm, which aresituated within the absorption spectrum of hemoglobin (Hb) andMyoglobin (Mb). Using two wavelengths allows for determinationof the changes in [HbO2] and deoxyhemoglobin ([HHb]) con-centration. The inter-optode distance (IOD) was fixed through theapparatus at 30, 35 and 40 mm. Positioning of the NIRS probe wasconsidered paramount to the study de Q4sign (Strangman et al., 2003;Boas et al. 2001; Mansouri et al. 2010) and great care was takenthat the probe was tangential to the curvature of the cranium andthat no contamination through ambient light was present. Fixationwas obtained through a custom-designed multi-density foam re-ceptacle which was secured using a system of straps to preventmovement during exercise. Location of the rIFC was determinedusing AF8 references from the electro-encephalic 10–20 interna-tional system and measured in duplicate. The probe position wasthen extensively marked on the skin using a surgical marker andwas also documented photographically for later reference to en-sure congruent placement in the following session. Data was ac-quired at 10 Hz. Concentrations were calculated using the standardform of the modified Beer–Lambert Law (Beer, 1851; Delpy et al.1988). For this calculation, the wavelength-specific extinctioncoefficients were extracted from Cope (1991) and the equallyspecific differential pathlength factors (DPF) were adopted fromDuncan et al. (1995). HbO2 concentration from the deepest source(IOD40, ca. 2 cm) has been previously reported to be the mostsensitive indicator of regional cerebral oxygenation changes and ofneural activity (Hoshi, Kobayashi and Tamura, 2001) and wastherefore selected as the primary outcome measure. Post-acqui-sition, the data was normalized to the 2-min rest period and thecontrol data was truncated to retain the same duration as cogni-tive task time completion in the exercise condition. Data wassubsequently reduced to 20 datapoints using a splineinterpolation.

2.6. Experimental procedure

Three sessions (one training session and two experimentalsessions) were conducted at standard (71 h) morning hours(from 8 to 12 pm) within close intervals (473 days). Participantswere instructed to abstain from any vigorous exercise for 24 h pre-session, to sleep at least seven hours in the night before eachsession and to avoid caffeinated drinks in the morning.

During the training session, subjects performed four blocks of80 trials of the cognitive task to prevent a potential learning effect.Additional blocks were completed if the participant did not fulfillthe following learning criteria: (a) RT intra-block variability below5%, (b) RT variability with the previous block below 5%, (c) meanRT inferior to 600 ms, and (d) response accuracy superior to 85%.Participants then performed a maximal aerobic power (MAP) teston the cycle ergometer. The resistance was automatically regulatedto ensure constant power output independent of pedal frequency.Power output was increased by 10 W every 30 s after a 4 min

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132

Fig. 1. Schematic drawing of an incongruent trial displayed at the top of the screen.In this example, participants had to respond by pressing the left response key ac-cording to the direction indicated by the central arrow.

C. Schmit et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 3


warm-up at light intensity (women: 70 W; men: 80 W). Oxygenconsumption (ml kg-1 min-1) and ventilatory output (L min-1)were recorded using the Fitmate Pro gas analyzer (COSMED,Miami, USA) validated by NiemaQ5 n et al. (2007). Cardiac frequencywas recorded by a Polar system (Polar RS800CX, Polar Electror Oy,Kempele, Finland). Voluntary exhaustion was defined as the pointwhere participants voluntarily stopped or when they could nolonger maintain a pedaling frequency above 50 rotations perminute (RPM) more than 10 s despite strong verbalencouragement.

The experimental sessions consisted of a control condition andan exercise condition (Fig. 2). The order of the sessions wascounterbalanced across participants. Each session began with twotraining blocks of 80 trials of the cognitive task (about 3 min).Following the installation of the NIRS and EMG sensors (about30 min), the measurement of Cox resting values was conducted.The recording of baseline values only began if Cox level was vi-sually stable during the last 2 min. Participants then performedthe cognitive task for twenty minutes1 without pedaling (controlcondition) or while pedaling at 85% of MAP until exhaustion (ex-ercise condition). The intensity (regulated independently frompedal frequency) was chosen to be intense enough to exceed thesecond ventilatory threshold (VT2) while allowing an exerciseduration long enough to obtain sufficient data to define two timeperiods: an initial period and a terminal period. In each condition,the initial period corresponded to the first hundred trials of thecognitive task performed by the participant. In the exercise con-dition, the terminal period corresponded to the last hundred trialsperformed just before exhaustion and was determined accordingto individual time to exhaustion. To obtain a matching terminalperiod in the control condition, the time frame corresponding tothe cessation of pedaling during the exercise condition was ex-tracted and the period calculated from there. The exercise condi-tion began after a 3-min warm-up at 50% of MAP. The intensitywas then increased every 20 s for 2 min until reaching 85% of MAP.To ensure that a steady-state of oxygen consumption was attained,a 90 s delay was imposed between the attainment of target in-tensity and the beginning of the cognitive task. Cox, heart rate(HR), pedaling frequency and electromyographic activity of theagonist muscles involved in the task were continuously monitored

from rest to exhaustion.

2.7. Psychological measurement

Participants were asked to provide a verbal rating of perceivedexertion (RPE) between each block of the cognitive task, using avisual Borg (6–20) scale (Borg, 1998). RPE was defined as the“perceived difficulty to exert at the same time the physical plus thecognitive tasks”. The next block followed immediately after par-ticipants’ RPE responses.

2.8. Data analysis

The EMG signals of the flexor pollicis brevis recorded duringeach trial were aligned to the onset of the imperative stimulus andthe onsets of the changes in activity were visually determined.EMG signals with a background activity superior to 10% of theburst peak were rejected (Rejection rate: 20%). RT for pure correcttrials–trials with no sign of EMG activation associated with theincorrect activation-was measured for each condition from onsetof the stimulus to the onset of the EMG involved in the response.RTs of less than 100 ms and RTs higher than 1500 ms (3% of thetotal number of trials) were considered anticipated responses andomissions, and were excluded from further analysis. The incorrectactivation trials were differentiated into two categories of trials:errors and partial EMG errors. Partial EMG errors were incorrectaction impulses, mostly undetected consciously, that were suc-cessfully corrected (Hasbroucq et al., 1999; Rochet et al., 2014). Theforce exerted by the non-required effector was not sufficient toelicit an error and was followed by a correct activation whichreached the response threshold. Trials containing a partial errorare of particular interest, since they indicate that although an errorwas about to be made, the nervous system was able to overcomeand provide the correct action. Error rate and partial EMG errorrate were calculated according to the number of trials after artifactrejection. Additionally, the correction rate which represents theefficiency of the nervous system to overcome and provide thecorrect action after incorrect activations was calculated. It corres-ponds to the number of partial EMG errors divided by the totalnumber of incorrect activations (partial EMG errors and errors).

For the exercise condition, the root mean square (RMS) fromEMG signal of the VL was calculated using a custom signal de-tection routine in Matlab between the onset and the end of eachburst recorded from the beginning of the cognitive task and until

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132

Fig. 2. Schematic representation of the experimental sessions which consisted of performing an Eriksen flanker task for 20 min without exercising (control condition) orwhile cycling at 85% of maximal aerobic power (MAP) (exercise condition). The initial period of exercise corresponds to the first hundred trials and the terminal period to thelast hundred trials in the exercise condition. These two time windows were matched with the control condition. Cox resting values were measured prior to any cognitiveand/or physical solicitation.

1 This corresponds to 1.5 times the best time-to-exhaustion performed duringpre-tests and thus is ensured to completely cover the cognitive task durationduring the exercise condition for all participants.



exhaustion. Results of each subject were reduced to 20 measure-ment points using a spline interpolation function in Matlab(R2012b, the MathWorks, Inc., Natick, MA) in order to depict themean evolution of VL muscular activation during the task.

Through the analyzes of the percentage of correct responses(conditional accuracy functions, CAF) and the magnitude of theinterference effect (delta curve) as a function of RT, the activation-suppression model provides a powerful framework to assessconflict resolution (for details see, van den Wildenberg et al.,2010). This model specifically allows for the assessment of boththe initial phase linked to an individual’s susceptibility to makingfast impulsive errors (early automatic response activation) and, thelater phase associated with the efficiency of the cognitive control(build-up of a top-down response suppression mechanism). Re-action time distribution was obtained using individual RTs “vin-centized” into four equal-size speed bins (quartiles) for CO and INtrials separately. The lack of data for three participants did notpermit a relevant vincentization, consequently they have not beentaken into account in this analysis. Delta plots were constructed bycalculating interference effect as a function of the response speed(average of difference between RT in IN and RT in CO trials for eachquartile). Curve accuracy functions (CAF) were constructed byplotting accuracy as a function of the response speed. The datapresented are the mean values of each set averaged acrossparticipants.

Separated ANOVAs were performed on each dependent vari-able (i.e., mean RT, partial EMG errors, errors and correction rates).The analyzes involved conditions (control vs. exercise) congruency(CO vs. IN) and periods (initial vs. terminal) as within-subjectfactors. In order to control for the effect of the order of the ex-perimental session, this was initially included as a between-sub-ject factor along with all its interaction terms with the otherpredictors in the analyzes. However, given that none of thesevariables reached significance (ps4 .10), the order and its inter-action terms were removed from the analyzes to optimize theparsimony of the models. An ANOVA including condition (controlvs. exercise) and period (initial vs. terminal) as within-subjectsfactors was performed on hemoglobin concentration measures todetermine whether cerebral oxygenation diverged. The analyzesconducted on the RMS of VL activation and RPE data recordedduring the exercise session only included period (initial vs. term-inal) as a within-subject factor. Exploratory Pearson correlationswere performed between [HbO2] concentration, cognitive perfor-mances (mean RT, error rate, partial error rate, correction rate) androot mean square (RMS) from EMG signal of the VL. Data pre-sented on these measures correspond to mean values averagedacross participants. The SPSS software (IBMs SPSSs Statistics 20)was employed for all analyzes. Planned comparisons were used inthe GLM as post-hoc analyzes when significant p-values (po .05)were found. Values are expressed as mean7standard deviation(SD).

3. Results

3.1. Time to task failure, heart rate and rating of perceived exertion

In the exercise condition, the strenuous exercise until exhaus-tion lasted 360743 s, which enabled participants to complete5.570.4 blocks of the cognitive task (i.e., average of 220 trials,ranged from 200 to 320 trials). Participants always stopped cyclingat the end of a block of the cognitive task, meaning they alwaysfully completed all the blocks they have started.

HR showed a main effect of condition (F(1,14)¼1507.95,po .001, ηp2¼ .99), a main effect of period (F(1,14)¼43.83, po .001,ηp2¼ .76) and an interaction between these two factors (F(1,14)¼

33.27, po .001, ηp2¼ .70). In the control condition, HR remainedstable from the initial (6774 beat per minute, bpm) to theterminal period (6673 bpm, F(1,14)o1, p¼ .73), but increasedfrom 16272 bpm to 17672 bpm in the exercise condition (F(1,14)¼152.12, po .001, ηp2¼ .92).

RPE results showed a main effect of period (F(1,14)¼387.18,po .001, ηp2¼ .96) and condition (F(1,14)¼450.92, po .001,ηp2¼ .97). An interaction between these two factors was observed(F(1,14)¼6.91, p¼ .02, ηp2¼ .33). The increase in RPE was greater inthe exercise condition (from 14.270.6 to 19.470.3) compared tothe control condition (6.570.2 to 8.970.7).

3.2. Reaction time

Results showed a main effect of congruency (F(1,14)¼29.72,po .001, ηp2¼ .68) with longer RT for IN trials (454713 ms) thanfor CO trials (41078 ms). No other main effect or interaction wassignificant.

3.3. Error rate

Results showed a main effect of congruency (F(1,14)¼60.18,po .001, ηp2¼ .81), a main effect of condition (F(1,14)¼9.54,po .01, ηp2¼ .40) as well as an interaction between these twofactors (F(1,14)¼11.74, po .01, ηp2¼ .45, Fig. 3). The frequency oferrors was lower during rest than when exercising for IN trials(Ctrl: 9.9371.14%; Exer: 18.1573.67%; po .01) but not for COtrials (Ctrl: 0.8670.53%; Exer: 1.2170.52%; p¼ .58) (Fig. 3). In-terestingly, the interaction between condition and period showeda trend (F(1,14)¼3.37, p¼ .08, ηp2¼ .19) and suggests that theevolution of accuracy differs between control and exercise condi-tions. In the terminal period, participants committed more errorsduring exercise (10.7373.84%) than at rest (5.3271.59%, po .01,ηp2¼ .40), whereas they had an equivalent error rate in the initialperiod (Exer: 8.3172.54%; Ctrl: 5.8871.41%; p4 .05).

3.4. Partial EMG error

Results showed a main effect of congruency (F(1,14)¼143.76,po .001, ηp2¼ .91), condition (F(1,14)¼11.32, po .01, ηp2¼ .45) andperiod (F(1,14)¼7.21, p¼ .01, ηp2¼ .34). The number of partial EMGerrors was increased for IN trials (36.9274.13%) compared to COtrials (10.6572.81%), exercise (28.0975.58%) compared to rest(19.4974.76%) and greater in the terminal period (25.8375.90%)than in the initial period (21.7474.26%). No interaction reached

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132

Fig. 3. Error rate (in percentage) during control (Ctrl, white bars) and exercise(Exer, grey bars) conditions for congruent (CO) and incongruent (IN) trials. Errorbars represent standard deviation.



significance. The interaction between condition and period did notreached significance (F(1;14)¼ .04; p¼ .84; ηp2¼ .003). All the re-sults for RT, error rate (accuracy) and partial EMG error are pre-sented in Table 2.

3.5. Correction rate

Correction rate trended towards significant interaction be-tween condition and period (F(1,14)¼4.12, p¼ .06, ηp2¼ .23, Fig. 4).In the initial period, the correction rate was equivalent at rest andduring exercise (F(1,14)¼0.82, p¼ .37, ηp2¼ .05). However, in theterminal period, participants were less capable to correct incorrectaction impulses during exercise (68.9273.42%) than in the controlcondition (78.0573.56%) (F(1,14)¼7.03, p¼ .01, ηp2¼ .33). Thisfinding suggests a deficit in cognitive control just before exhaus-tion, incorrect action impulses were not corrected effectively andmore errors were committed.

3.6. Distributional analysis

Reaction time distributions were submitted to an ANOVA in-volving condition (control vs. exercise), congruency (CO vs. IN),period (initial vs. terminal) and quartile (Q1, Q2, Q3, Q4) as within-subjects factors. Results confirmed the main effect of congruencypreviously observed on mean RT (F(1, 11)¼73.07, po .001,ηp2¼ .87). More interestingly, the analysis showed an interactionbetween condition and quartile (F(3,33)¼5.16, po .01, ηp2¼ .32)which revealed that exercise differently affects RT performance as

a function of the response speed (Fig. 5A). A beneficial effect ofexercise was actually observed for the first quartile (Q1: "27 ms,po .01), and the second quartile (Q2: "21 ms, po .05) and dis-appeared for the last two quartiles (Q3: "11 ms, p¼ .19; Q4,þ13 ms, p¼ .10).

According to van den Wildenberg et al. (2010), the magnitudeof interference as RT lengthened is associated with the efficiencyof the cognitive control (build-up of a top-down response sup-pression mechanism). Then, a second analysis focused on deltaplot slopes was conducted to examine whether exercise alteredthe magnitude of the interference. The analysis involved condition(control vs. exercise), period (initial vs. terminal) and quartile (Q1,Q2, Q3, Q4) as within-subject factors. Results showed that themagnitude of interference did not fluctuate as RT lengthened,except during the initial period of exercise (Fig. 5B). In this period,the interference decreased from 64 ms (79 ms) in the first quar-tile to 25 ms (712 ms) in the last quartile (F(1,11)¼14.19, po .01)suggesting that non-exhausting intense exercise enhances cogni-tive control. It is noteworthy that no sign of any deficit in cognitivecontrol was observed in the terminal period of exercise whenexhaustion was about to occur (F(1,11)¼0.07, p¼ .79).

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132

Table 2Mean reaction times, accuracies and partial errors per condition and period in theEriksen flanker task.

Variables Control ExerciseInitial Terminal Initial Terminal

CO RT (ms) 411734 414743 406740 410749Acc (%) 1.772.3 0.771.7 171.9 0.772.1PE (%) 776.8 8.276.5 12.879.1 14.7710.3

IN RT (ms) 454750 461746 447763 456776Acc (%) 1073.7 9.974.9 15.778.2 20.6714.2PE (%) 27.6714.8 35.2714.5 39.6714 45.3717.3

Results are presented as the mean group 7 SD.Notes. SD¼standard deviation; CO¼congruent trials; IN¼ incongruent trials;RT¼reaction time; Acc¼accuracy; PE¼partial error.

Fig. 4. Correction rate (in percentage) during rest (Ctrl, white bars) and exercise(Exer, grey bars) conditions for the initial and terminal periods. Error bars representstandard deviation.

Fig. 5. (A) Cumulative density functions as a function of reaction time (RT) duringrest (Ctrl, empty symbols) and exercise (Exer, full symbols) conditions. (B) Deltaplots of RT illustrating the magnitude of the interference (in ms) as a function of RTduring rest (Ctrl, circle) and exercise (Exer, triangle) conditions for the initial(empty symbols) and terminal (full symbols) periods.



3.7. Conditional accuracy functions (CAF)

An ANOVA involving condition (control vs. exercise), con-gruency (CO vs. IN), period (initial vs. terminal) and quartile (Q1,Q2, Q3, Q4) as within-subject factors was conducted on error rates(Fig. 6). Results confirmed the main effects of condition (F(3,14)¼9.78, po .01, ηp2¼ .41), congruency (F(3,14)¼40.35, po .001,ηp2¼ .74) and the interaction between the two (F(3,14)¼11.14,po .01, ηp2¼ .44) previously observed on mean error rate. In linewith the activation–suppression model (Ridderinkhof, 2002), theinteraction between congruency and quartile, which illustratedthe strength of the automatic response triggered by the flankers,was significant (F(3, 42)¼25.02, po .001, ηp2¼ .64). The inter-ference was more pronounced for the first quartile than for thesecond quartile (Q1: 31 ms vs. Q2: 12 ms, po .001) and for thesecond quartile compared to the third quartile (Q2: 12 ms vs. Q3:8 ms, po .01), whereas the interference was equivalent for the lasttwo quartiles (Q3: 8 ms vs. Q4: 6 ms, p¼ .23). The interaction be-tween condition, congruency and period (F(1,14)¼3.89, p¼ .07,ηp2¼ .22) and the interaction between condition, congruency and

quartile (F(3,14)¼2.71, p¼ .07, ηp2¼ .16) tended to be significant.A second series of analyzes, focusing on the first quartile of the

CAF, was conducted to examine whether exercise alters the rapidresponse impulse (van den Wildenberg et al., 2010). The analysiscarried out on the initial period of exercise did not reveal an in-teraction between condition and congruency (F(1,14)¼1.05,p¼ .32, Fig. 6A). However, in the terminal period of exercise, therewas a significant interaction (F(1,14)¼12.68, po .01, ηp2¼ .49). Justbefore exhaustion, participants committed about 15% more errorsthan at rest in IN trials (Fig. 6B) while the accuracy rate in CO trialsremained unchanged.

3.8. Cerebral oxygenation

An interaction between condition and period was observed onCox fluctuations (F(1,14)¼5.98, po .001, ηp2¼ .30) (Fig. 7). Duringthe control condition, the time spent on the task did not impactCox (F(1,14)¼0.71, p¼ .98, ηp2¼ .04). During exercise, ∆[HbO2]linearly (linearity, po .05) decreased from 2.7170.05 mmol cm inthe initial period to 0.2170.06 mmol.cm in the terminal period

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132

Fig. 6. Conditional accuracy function (CAF) representing the percentage of accuracy for congruent (CO, empty symbols) and incongruent (IN, full symbols) trials as a functionof reaction time (RT) during control (Ctrl, circle) and exercise (Exer, triangle) conditions for the initial (A) and terminal (B) periods.

Fig. 7. Changes from resting values (baseline) in cerebral oxyhemoglobin [HbO2] during control (Ctrl, circles) and exercise (Exer, triangles) conditions. Error bars representstandard deviation.



(F(1,14)¼9.16, po .001, ηp2¼ .40)2. In the terminal period, Cox waslower while exercising than in the control condition(1.8870.1 mmol cm, F(1,14)¼5.64, p¼ .03, ηp2¼ .30).

[HbO2] levels recorded per condition and period negativelycorrelated with error rate (r¼" .40; p¼ .001) and, more specifi-cally, with error rate on IN trials (r¼" .39; po .01). [HbO2] levelsalso negatively correlated with partial errors both on CO (r¼" .38;po .01) and IN trials (r¼" .26; po .05). Mean RT (r¼ .04; p¼ .79)and the correction rate (r¼ .17; p¼ .19) were not associated tochanges in [HbO2] levels.

3.9. Vastus lateralis activation

A significant main effect of time was observed on the VL ac-tivity (F(1,14)¼3.92, po .001, ηp2¼ .23), which illustrates an in-creased activation of the muscle throughout exercise duration(Fig. 8). Furthermore, it is interesting to note that the RMS valuesare significantly and negatively correlated with exercise [HbO2](r¼" .23; po .05).

4. Discussion

This study aimed at investigating concomitant changes incognitive control and Cox in the prefrontal area during strenuousexercise performed until exhaustion. In an initial period, Cox re-corded from the rIFC remained unchanged by intense exercise.Also, the more pronounced drop-off of the delta curve suggestedthat the cognitive processes, underpinning selective response in-hibition, are fully efficient in the first part of the exercise bout.Throughout intense exercise and until exhaustion Cox linearlydecreased but without falling below baseline values. In the term-inal period, no sign of deficit in selective response inhibition wasobserved. However, individual’s susceptibility to making fast im-pulsive errors increased and less efficient online correction of in-correct activation was observed just before exhaustion. The mainfindings of this study are that (i) cognitive functioning evolvesduring exercise, (ii) intense exercise does not systematically impaircognitive performances, (iii) selective response inhibition effi-ciency and PFC Cox do not follow a similar dynamic, (iv) the

propensity to commit impulsive errors increases and online cor-rection of incorrect activation are disrupted near exhaustion, and(v) Cox patterns suggest a decline in hyperfrontality instead of ahypofrontality.

4.1. Intense exercise and cognitive performance

The experimental as well as theoretical literature on the cog-nitive effect of intense exercise converges towards an impairmentof cognitive performances (Ando et al., 2005; Chmura et al., 1994;Chmura and Nazar, 2010; Cooper, 1973; Dietrich and Audiffren,2011; McMorris et al., 2008; Yerkes and Dodson, 1908). By in-vestigating changes in cognitive functioning through distributionalanalyzes, the present study highlights a facilitating effect of in-tense exercise on cognitive control. This effect was localized in theinitial part of the exercise bout. Both RT performances and selec-tive response inhibition were indicative of this improvement.More precisely, the exercise-related speeding effect focused on thefirst two quartiles of the RT distribution, i.e. the fastest RT. Thisdoes not appear surprising since faster RT have been consistentlyreported from several meta-analyzes and integrative reviews(Brisswalter et al., 2002; McMorris and Hale, 2012, Tomporowski,2003). In contrast, the benefit of intense exercise on selective re-sponse inhibition constitutes an innovative result. Concretely, thesteep negative slope of the delta plots indicates an exercise-relatedlower interference effect compared to rest. The activation-sup-pression model of Ridderinkhof (2002) proposes that such pro-nounced leveling-off in the delta curve is indicative of a greaterability to suppress the automatic response generated by task-ir-relevant aspect of the stimulus. The fact that the correction rateremained constant during the initial part of the exercise suggeststhat online control mechanisms, involved in the correction of in-correct activation, are fully efficient. Together, these findings revealthat the facilitating effect usually reported during moderate ex-ercise can also occur in the first moments of intense exercise.

A cognitive facilitation during intense exercise is not in dis-crepancy with the main theories on exercise and cognition. Spe-cifically, the catecholamine theory predicts cognitive functioningimpairment above the “catecholamine threshold” i.e. a pivotalpoint into the exercise-induced increase in adrenaline, nora-drenaline and dopamine (Cooper, 1973; McMorris et al., 2008).During steady exercise, since the level of monoamines increasesover time regardless the intensity (Chmur Q6a et al., 1997), a certainamount of time is necessary before overreaching this threshold.Accordingly, the first stages of our exercise bout may have spared

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132

Fig. 8. Electromyographic root mean square (in mV) activity of the vastus lateralis muscle during the time-to-exhaustion cycling test. Error bars represent standard errors.

2 During exercise, the same pattern was observed in total hemoglobin([HbO2]þ[HHb]) which is considered another index of neural activity (e.g., Perrey,2008). Total hemoglobin decreased from 5.4970.15 mmol cm in the initial period to0.697 0.36 mmol cm in the terminal period (F(1,14)¼28.35, po .001, ηp2¼ .67).



central processes from neural noise, while simultaneously arous-ing central nervous system and benefiting RT. The transient hy-pofrontality theory (Dietrich, 2003; Dietrich and Audiffren, 2011)purports another perspective, in which exercise leads to re-arrangement of neural resources within the brain. Given the lim-ited cerebral resources, exercise would lead to their redistributionfrom the brain regions that are not directly involved in the man-agement of exercise such as the PFC to motor areas. Our fatiguingexercise required an increasing magnitude of motor cortex outputto increase the firing rate of motor neurons in order to compensatemuscle fiber fatigue, as supported by the recorded increase in RMSvalue of the VL muscle. Instead of exercise intensity per se, main-taining the steady power output over time may thus have pro-gressively acted in favor of a PFC down-regulation.

4.2. Exhausting exercise and cognitive performance

Based on the distribution-analytical technique and the deltaplot analysis, the present results show that, when exhaustion wasabout to occur, selective inhibition processes remained unalteredand were thus not responsible for inferior behavioral perfor-mances. The delta curve indeed highlights that the selective in-hibition is as efficient as in the control condition. Nevertheless, thenumber of incorrect response activations – including both overtand partials EMG errors – increased. The analyzes of the percen-tage of correct responses (CAF) showed that, when exhaustion wasabout to occur, the individual’s susceptibility to produce incorrectresponses increases. In other words, exhaustion state increases thestrength of the automatic response capture activated by irrelevantinformation and more overt errors are committed.

Interestingly, the recording of the EMG of the FPB muscle in-volved in the cognitive task allowed us to identify partial errortrials i.e. incorrect action impulses that were detected and suc-cessfully corrected. Accordingly, partial errors provide a directmeasure of the effectiveness of the control mechanism involved inthe suppression of the activation of an incorrect response. Thepresent results showed that, in the terminal period, participantswere less capable to correct incorrect action impulses during ex-ercise than in the control condition and more overt errors werecommitted. This finding suggests that, near exhaustion, the onlinecorrection mechanism is disrupted and the nervous system seemsnot able to overcome the incorrect activation and provide thecorrect action. Since the participants aimed at pursuing exercise aslong as they could, it is possible that, near exhaustion, motor task-related regulations (cardio-respiratory, velocity, coordination,power output) became such imperative that they were managed inpriority at the expense of cognitive task-related regulations (seeSection 4.4).

By dissociating our exhausting exercise into analysis periods,we also investigate a new perspective of exercise-cognition stu-dies. This perspective, actually based on a fatigue-induced re-organization, may help to clarify current inconsistencies in theliterature. Concretely, it rationalizes why a sustained cognitivesolicitation during last stages of a 65-min exercise bout may revealdiminished performance (Dietrich and Sparling, 2004) while40 min of an intermittent assessment does not (Lambourne et al.,2010), in spite of similar moderate intensities and cognitive tasks.According to this proposal, the impaired cognitive performanceassociated with intense exercise do not appear illogical. Indeed,fatigue development is obviously quicker at such higher in-tensities. Considering this accumulation of fatigue, other mod-erators should also be taken into account. Fitness level, for ex-ample, may explain the better (Chang et al., 2012) and steadier(Labelle et al., 2012) cognitive performance of trained participants.Beyond this, we would like to encourage the general idea oftemporal differentiation within data sets to better understand

integrated fatigue development.

4.3. Cerebral oxygenation and cognitive performance

In the present study, Cox recorded during exercise from therIFC was at first at a similarly elevated level as in the controlcondition before linearly declining until exhaustion. This type of[HbO2] decrease is common during exhausting exercise and inaccordance with previous reports from PFC NIRS-monitoring stu-dies (for details see Ekkekakis, 2009).

More particularly, we found that the Cox level was reducedduring the terminal period of exercise compared to the controlcondition where the cognitive task was conducted at rest. In spiteof this decline, we observed that the implementation of selectiveresponse inhibition remained fully efficient (comparable to thescore of the control condition). This is intriguing since rIFC activityis an important component in inhibition processes (Aron et al.,2004) and a debilitative cognitive effect might thus be expectedfrom its down-regulation. This report is not isolated though. Arecent study observed a similar discrepancy: cognitive perfor-mance improved at moderate intensity (60% VO2̇ ) in the absence ofany changes in Cox values (Ando et al., 2011). This might lead tothe suggestion that an uncoupling of Cox level in PFC areas andcorresponding cognitive processes may be happening. This type ofuncoupling would not ineluctably hamper, but could maintain PFCfunctionality. As an explanation, the PFC may preserve its meta-bolic activity by increasing oxygen extraction from arterial vesselsto compensate for reduced perfusion (Nybo and Secher, 2004). It isalso possible that the Eriksen flanker task was not demandingenough to elicit observable behavioral effects from reduced Coxlevel.

Exercise-induced hyperventilation is considered to be the mainmechanism for the lowering of cerebral blood flow and, in turn,Cox level (Ogoh and Ainslie, 2009). In our study, ventilatorymuscle fatigue may have led to this progressive drift into venti-lation and hypocapnia. In spite of this process, the [HbO2] con-centration never reached values lower than baseline (i.e. a statethat could be characterized as hypofrontal). Since [HbO2] levelconsistently remained positive, our Cox pattern rather supportsthe decline of a hyperfrontality state. This contrasts with the re-ticular-activating hypofrontality theory (Dietrich and Audiffren,2011) but not with some of its principles. Specifically, whenviewed in light of the redistribution of cerebral resources, somefindings may be considered as a support to the theory. Indeed, onepossible explanation is that PFC was progressively inhibited as aside-effect of fatigue development to favor activity in motor areas,as supported by the Cox-RMS correlation. In this case, both thecorrelations between [HbO2] and error rates, between [HbO2] andpartial error rates, and CAF results near exhaustion support thehypothesis of a reallocation, since impulsive errors relate to ac-tivity of the pre-supplementary motor area (Forstmann et al.,2008).

4.4. Rationalize the relation between exercise and cognitiveperformance

Our results reinforce the idea of an interaction between ex-ercise and cognition for the complete duration of an exercise bout.This interference has previously been proposed using strength(Lorist et al., 2002; Schmidt et al., 2009) and aerobic exercises(Marcora et al., 2009; McCarron et al., 2013). Accordingly, we as-sume that behavioral performance relative to cognitive tasks ispunctual and systemic and depends on the constraints supportedby the subject at a given time. This idea of a dynamical cognitivecontrol is supported from several perspectives.

Marcora (2008, 2 Q7009) proposes a psychobiological model of

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132



exercise, within which the anterior cingulate cortex (ACC) appearsas the keystone of both exercise and cognitive parameters. ACC isknown to be involved in cognitive functioning (Carter et al., 1998),the pain matrix (Peyron et al., 2000), perceived effort (Williamsonet al., 2006) and effort-related decision-making. Regarding ourprotocol and its demands, a hyper-solicitation of the ACC over timemay compromise its efficacy to deal with interfering stimuli.

The insular cortex and hypothalamus are other brain areas thatare increasingly activated with fatigue development (Meynielet al., 2013). In response to exercise duration and increasing bodyafferences, it is possible that these regions act to reduce basalganglia activation. Such inhibition would prevent the subject fromexperiencing untolerable perceived effort or any excessive home-ostasis disruption, but would be enforced at the expense of theoverall performance. Indeed, basal ganglia (specifically the ventralstriatum) activation determines both cognitive and motor efforts(Schmidt et al., 2012). Near exhaustion the subject may thus, vo-luntarily or not, opt for a facilitating strategy leading him to pro-gressively act on the basis of impulsive activations rather than onthe basis of high-order processes.

The neuro-hormonale rationale of the “catecholamines hy-pothesis” may also determine the way participants respond to acognitive task during exercise (McMorris et al., 2009). Due to therole of monoamines in glycolysis, lipolysis and cardio-respiratoryregulation (Borer, 2003), sustained exercise induces increases inadrenaline, noradrenaline and dopamine irrespective of its in-tensity (Chmura et al., 1997). Such accumulation may progressivelylead to overreach the “catecholamine threshold” that would in-duce neural noise and contributes to the cessation of exercise-in-duced cognitive facilitation.

5. Conclusion

In conclusion, this study is innovative in that changes in cog-nitive performances during a steady exercise were characterized.The benefit of intense exercise on selective response inhibitionconstitutes an original result. Moreover, the use of the distribu-tion-analytical technique highlighted that, when exhaustion wasabout to occur, selective inhibition processes remained unaltered.Despite this, individual’s susceptibility to making fast impulsiveerrors increased and less efficient online correction of incorrectactivation was observed, suggesting that the online correctionmechanism is disrupted. Interestingly, the dynamical pattern ofselective response inhibition efficiency did not follow the samepattern as [HbO2], letting Cox-related explanations of cognitivefunctioning during exercise uncertain. These results reinforce theidea of a complex interaction between exercise and cognition andinclude fatigue stressors as a determinant component into cogni-tive performances.

Declaration of interest

The authors report no conflict of interest. The authors alone areresponsible for the content and writing of the paper.

Uncited refereQ8 nces

Chmura et al. (1998), Davranche and Pichon (2005), Meeusenand De Meirleir (1995), Subudhi et al. (2007), Yamashita et al.(2001)

Acknowledgements

The authors would like to thank Pelle Bernhold for his con-structive help in the study reasoning, and Laura Gray for Englishrevisions. The Brainbike was provided by Hitech Fitness, Valbonne,France, although the authors have no conflict of interest to report.

References

Ando, S., Kimura, T., Hamada, T., Kokubu, M., Moritani, T., Oda, T., 2005. Increase inreaction time for the peripheral visual field during exercise above the venti-latory threshold. Eur. J. Appl. Physiol. 94, 461–467.

Ando, S., Kokubu, M., Yamada, Y., Kimura, M., 2011. Does cerebral oxygenation affectcognitive function during exercise? Eur. J. Appl. Physiol. 111, 1973–1982.

Aron, A.R., Robbins, T.W., Poldrack, R.A., 2004. Inhibition and the right inferiorfrontal cortex. Trends Cogn. Sci. 8, 170–177.

Aron, A.R., Robbins, T.W., Poldrack, R.A., 2014. Inhibition and the right inferiorfrontal cortex: one decade on. Trends Cogn. Sci. 18, 177–185.

Beer, A., 1851. Versuch der Absorptions-Verhältnisse des Cordierites für rothes Lichtzu bestimmen. Ann. Phys. Chem. 84, 37–52.

Bhambhani, Y., Malik, R., Mookerje, S., 2007. Cerebral oxygenation declines at ex-ercise intensities above the respiratory compensation threshold. Respir. Physiol.Neurobiol. 156, 196–202.

Boas, D.A., Gaudette, T., Strangman, G., Cheng, X., Marota, J.J.A., Mandeville, J.B.,2001. The Accuracy of Near Infrared Spectroscopy and Imaging during FocalChanges in Cerebral Hemodynamics. NeuroImage 13 (1), 76–90.

Borer, K.T., 2003. Exercise Endocrinology. Human Kinetics, Champaign, IL.Borg, G., 1998. Borg’s Perceived Exertion and Pain Scales. Human Kinetics, Cham-

paign, IL.Brisswalter, J., Collardeau, M., Arcellin, R., 2002. Effects of acute physical exercise

characteristics on cognitive performance. Sports Med. 32, 555–566.Carter, C.S., Braver, T.S., Barch, D.M., Botvinick, M.M., Noll, D., Cohen, J.D., 1998.

Anterior cingulate cortex, error detection, and the online monitoring of per-formance. Science 280 (5364), 747–789.

Chang, Y.K., Labban, J.D., Gapin, J.I., Etnier, J.L., 2012. The effect of acute exercise oncognitive performance: a meta-analytic review. Brain Res. 1453, 87–101.

Chmura, J., Nazar, K., Kacuba-Uscilko, H., 1994. Choice reaction time during gradedexercise in relation to blood lactate and plasma catecholamine thresholds. Int. J.Sports Med. 15 (4), 172–176.

Chmura, J., Krysztofiak, H., Ziemba, A.W., Nazar, K., Kacuba-Uscilko, H., 1998. Psy-chomotor performance above and below the bloos lactate threshold. Eur. J.Appl. Physiol. 77, 77–80.

Chmura, J., Nazar, K., 2010. Parallel changes in the onset of blood lactate accumu-lation (OBLA) and threshold of psychomotor performance deterioration duringincremental exercise after training in athletes. Int. J. Psychophysiol. 75,287–290.

Cooper, C.J., 1973. Anatomical and physiological mechanisms of arousal with spe-cific reference to the effects of exercise. Ergonomics 16, 601–609.

Cope, M., 1991. The Development of a Near-Infrared Spectroscopy System and ItsApplication for Noninvasive Monitoring of Cerebral Blood and Tissue Oxyge-nation in the Newborn Infant. Univ. College London, London (Ph.D. thesis).

Davranche, K., Pichon, A., 2005. Critical flicker frequency threshold increment afteran exhausting exercise. J. Sport Exerc. Psychol. 27 (4), 515–520.

Davranche, K., Hall, B., McMorris, T., 2009a. Effect of acute exercise on cognitivecontrol required during an Eriksen flanker task. J. Sport Exerc. Psychol. 31,628–639.

Davranche, K., Paleresompoulle, D., Pernaud, R., Labarelle, J., Hasbroucq, T., 2009b.Decision making in elite white-water athletes paddling on a kayak ergometer. J.Sport Exerc. Psychol. 31 (4), 554–565 (2009 Aug).

Davranche, K., McMorris, T., 2009. Specific effects of acute moderate exercise oncognitive control. Brain Cogn. 69 (3), 565–570.

Delp, M.D., Armstrong, R.B., Godfrey, D.A., Laughlin, M.H., Ross, C.D., Wilkerson, M.K., 2001. Exercise increases blood flow to locomotor, vestibular, cardior-espiratory and visual regions of the brain in miniature swine. J. Physiol. 533,849–859.

Delpy, D.T., Cope, M., Zee, P., van der., Arridge, S., Wray, S., Wyatt, J., 1988. Esti-mation of optical pathlength through tissue from direct time of flight mea-surements. Phys. Med. Biol. 33, 1433–1442.

De Pauw, K., Roelands, B., Cheung, S.S., de Geus, B., Meeusen, R., 2013. Guidelines toclassify subject groups in sport-science research. Int. J. Sport Physiol. Perform. 8(2), 111–122.

Dietrich, A., 2003. Functional neuroanatomy of altered states of consciousness: thetransient hypofrontality hypothesis. Conscious. Cogn. 12, 231–256.

Dietrich, A., 2009. The cognitive neuroscience of exercise: the transient hypo-frontality theory and its implications for cognition and emotion. In: McMorris,T., Tomporowski, P.D., Audiffren, M. (Eds.), Exercise and Cognitive Function.Wiley, New-York, pp. 69–90.

Dietrich, A., Audiffren, M., 2011. The reticular-activating hypofrontality (RAH)model of acute exercise. Neurosci. Biobehav. Rev. 35, 1305–1325.

Dietrich, A., Sparling, P.B., 2004. Endurance exercise selectively impairs prefrontal-dependent cognition. Brain Cogn. 55, 516–524.

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132



Duncan, A., Meek, J.H., Clemence, M., Elwell, C.E., Tyszczuk, L., Cope, M., Delpy, D.,1995. Optical pathlength measurements on adult head, calf and forearm andthe head of the newborn infant using phase resolved optical spectroscopy.Phys. Med. Biol. 40 (2), 295.

Eriksen, B.A., Eriksen, C.W., 1974. Effects of noise letters upon the identification of atarget letter in a non search task. Percept. Psychophys. 16, 143–149.

Ekkekakis, P., 2009. Illuminating the black box: investigating prefrontal corticalhemodynamics during exercise with near-infrared spectroscopy. J. Sport Exerc.Psychol. 2009, 505–553.

Forstmann, B.U., van den Wildenberg, W.P.M., Ridderinkhof, K.R., 2008. Neuralmechanisms, temporal dynamics, and individual differences interference con-trol. J. Cogn. Neurosci. 20, 1854–1865.

Gross, P.M., Marcus, M.L., Heistad, D.D., 1980. Regional distribution of cerebral bloodflow during exercise in dogs. J. Appl. Physiol. 48, 213–217.

Hasbroucq, T., Possamaï, C.-A., Bonnet, M., Vidal, F., 1999. Effect of the irrelevantlocation of the response signal on choice reaction time: an electromyographicstudy in humans. Psychophysiology 36 (4), 522–526.

Hazeltine, E., Poldrack, R., Gabrieli, J.D., 2000. Neural activation during responsecompetition. J. Cogn. Neurosci. 12 (suppl. 2), 118–129.

Holschneider, D.P., Maarek, J.-M.I., Yang, J., Harimoto, J., Scremin, O.U., 2003.Functional brain mapping in freely moving rats during treadmill walking. J.Cereb. Blood Flow Metab. 23, 925–932.

Hoshi, Y., Kobayashi, N., Tamura, M., 2001. Interpretation of near-infrared spec-troscopy signals: a study with a newly developed perfused rat brain model. J.Appl. Physiol. 90, 1657–1662.

Joyce, J., Smyth, P.J., Donnelly, A.E., Davranche, K., 2014. The Simon task and aging:does acute moderate exercise influence cognitive control? Med. Sci. SportsExerc. 46 (3), 630–639.

Kemppainen, J., Aalto, S., Fujimoto, T., Kalliokoski, K.K., Langsjo, J., Oikonen, V.,Rinne, J., Nuutila, P., Knuuti, J., 2005. High intensity exercise decreases globalbrain glucose uptake in humans. J. Physiol. 568, 323–332.

Labelle, V., Bosquet, L., Mekary, S., Bherer, L., 2012. Decline in executive controlduring acute bouts of exercise as a function of exercise intensity and fitnesslevel. Brain Cogn. 81, 10–17.

Lambourne, K., Audiffren, M., Tomporowski, P.D., 2010. Effects of acute exercise onsensory and executive processing tasks. Med. Sci. Sport Exerc., 1396–1402.

Lambourne, K., Tomporowski, P.D., 2010. The effect of exercise-induced arousal oncognitive task performance: A meta-regression analysis. Brain Res. 1341, 12–24.

Lorist, M.M., Kernell, D., Meijman, T.F., Zijdewind, I., 2002. Motor fatigue and cog-nitive task performance in humans. J. Physiol. 545 (1), 313–319.

Mansouri, C., L’Huillier, J.P., Kashou, N.H., Humeau, A., 2010. Depth sensitivityanalysis of functional near-infrared spectroscopy measurement using three-dimensional Monte Carlo modelling-based magnetic resonance imaging. LasersMed. Sci. 25 (3), 431–438.

Marcora, S.M., Staiano, W., Manning, V., 2009. Mental fatigue impairs physicalperformance in humans. J. Appl. Physiol. 106, 857–864.

McCarron, J., Hodgson, T.L., Smith, M.F., 2013. Brain drain: evaluating tQ9 he impact ofincreased cognitive load during self-paced running performance, Br. J. SportsMed. (2013).

McMorris, T., Collard, K., Corbett, J., Dicks, M., Swain, J.P., 2008. A test of the ca-techolamines hypothesis for an acute exercise–cognition interaction. Pharma-col. Biochem. Behav. 89, 106–115.

McMorris, T., Davranche, K., Jones, G., Hall, B., Corbett, J., Minter, C., 2009. Acuteincremental exercise, performance of a central executive task, and sym-pathoadrenal system and hypothalamic-pituitary-adrenal axis activity. Int. J.Psychophysiol. 73, 334–340.

McMorris, T., Sproule, J., Turner, A., Hale, B.J., 2011. Acute, intermediate intensityexercise, and speed and accuracy in working memory tasks: a meta-analyticalcomparison of effects. Physiol. Behav. 102, 401–408.

McMorris, T., Hale, B.J., 2012. Differential effects of differing intensities of acuteexercise on speed and accuracy of cognition: a meta-analytical investigation.Brain Cogn. 80, 338–351.

Meeusen, R., De Meirleir, K., 1995. Exercise and brain neurotransmission. SportsMed. 20, 160–188.

Meyniel, F., Sergent, C., Rigoux, L., Daunizeau, J., Pessiglione, M., 2013. Neuro-computational account of how the human brain decides when to have a break.Proc. Natl. Acad. Sci. USA 110, 2641–2646.

Nybo, L., Secher, N.L., 2004. Cerebral perturbations provoked by prolonged exercise.Prog. Neurobiol. 72, 223–261.

Ogoh, S., Ainslie, P.L., 2009. Cerebral blood flow during exercise: mechanisms ofregulation. J. Appl. Physiol. 107, 1310–1380.

Perrey, S., 2008. Non-invasive NIR spectroscopy of human brain function duringexercise. Methods 45 (4), 289–299.

Peyron, R., Laurent, B., Garcïa-Larrea, L., 2000. Functional imaging of brain re-sponses to pain. A review and meta-analysis. Clin. Neurophysiol. 30, 263–288.

Pontifex, M.B., Hillman, C.H., 2007. Neuroelectric and behavioral indices of inter-ference control during acute exercise. Clin. Neurophysiol. 118, 570–580.

Ridderinkhof, K.F., 2002. Micro- and macro-adjustments of task set: activation andsuppression in conflict task. Psychol. Res. 66, 312–323.

Ridderinkhof, K.R., van den Wildenberg, W.P.M., Wijnen, J., Burle, B., 2004. Re-sponse inhibition in conflict tasks is revealed in delta plots. In: Posner, M. (Ed.),Cognitive Neuroscience of Attention. Guilford Press, New York.

Rochet, N., Spieser, L., Casini, L., Hasbroucq, T., Burle, B., 2014. Detecting and cor-recting partial errors: evidence for efficient control without conscious access.Cogn. Affect. Behav. Neurosci. 14 (3), 970–982.

Rooks, C.R., Nathaniel, J.T., McCully, K.K., Dishman, R.K., 2010. Effects of incrementalexercise on cerebral oxygenation measured by near-infrared spectroscopy: asystematic review. Prog. Neurobiol. 92, 134–150.

Rupp, T., Perrey, S., 2008. Prefrontal cortex oxygenation and neuromuscular re-sponses to exhaustive exercise. Eur. J. Appl. Physiol. 102, 153–163.

Schmidt, L., Cléry-Melin, M.-L., Lafargue, G., Valabrègues, R., Fossati, P., Dubois, B.,Pessiglione, M., 2009. Get aroused and be stronger: emotional facilitation ofphysical effort in the human brain. J. Neurosci. 29 (30), 9450–9457.

Schmidt, L., Lebreton, M., Cléry-Melin, M.-L., Daunizeau, J., Pessiglione, M., 2012.Neural mechanisms underlying motivation of mental versus physical effort.Plos Biol. 10 (2).

Shibuya, K., Tanaka, J., Kuboyama, N., Murai, S., Ogaki, T., 2004a. Cerebral cortexactivity during supramaximal exhaustive exercise. Respir. Physiol. Neurobiol.140, 165–172.

Shibuya, K., Tanaka, J., Kuboyama, N., Ogaki, T., 2004b. Cerebral cortex activityduring intermittent supramaximal exercise. J. Sports Med. Phys. Fit. 44,215–219.

Strangman, G., Franceschini, M.A., Boas, D.A., 2003. Factors affecting the accuracy ofnear-infrared spectroscopy concentration calculations for focal changes inoxygenation parameters. NeuroImage 18, 865–879.

Subudhi, A.W., Dimmen, A.C., Roach, R.C., 2007. Effects of acute hypoxia on cerebraland muscle oxygenation incremental exercise. J. Appl. Physiol. 103, 177–183.

Tashiro, M., Itoh, M., Fujimoto, T., Fujiwara, T., Ota, H., Kubota, K., Higuchi, M.,Okamura, N., Ishi, K., Berecski, D., Sasaki, H., 2001. 18F FDG PET mapping ofregional brain activity in runners. J. Sports Med. Phys. Fit. 41, 11–17.

Timinkul, A., Kato., M., Omori., T., Deocaris., C.C., Ito., A., Kizuka, T., et al., 2008.Enhancing effect of cerebral blood volume by mild exercise in healthy youngmen: a near- infrared spectroscopy study. Neurosci. Res. 61, 242–248.

Timinkul, A., Omori, T., Wanpen, S., Yamada, Y., Soya, H., 2010. High frequency brainwaves increase depending upon exercise intensity and associated with cerebralblood volume: a simultaneous NIRS-EEG study. J. Med. Tech. Phys. Ther. 22 (1),81–91.

Tomporowski, P.D., 2003. Effects of acute bout of exercise on cognition. Acta Psy-chol. 112, 297–324.

Toronov, V., Webb, A., Choi, J.H., Wolf, M., Michalos, A., Gratton, E., Hueber, D., 2001.Investigation of human brain hemodynamics by simultaneous near-infraredspectroscopy and functional magnetic resonance imaging. Med. Phys. 28.

van den Wildenberg, W.P.M., Wylie, S.A., Forstmann, B.U., Burle, B., Hasbroucq, T.,Ridderinkhof, K.R., 2010. To head or heed? Beyond the surface of selective ac-tion inhibition: a review. Front. Hum. Neurosci. 4, 1–13.

Vissing, J., Anderson, M., Diemer, N.H., 1996. Exercise-induces changes in localcerebral glucose utilization in the rat. J. Cereb. Blood Flow Metab. 16, 729–736.

Williamson, J.W., Fadel, P.J., Mitchell, J.H., 2006. New insights into central cardio-vascular control during exercise in humans: a central command update. Exp.Physiol. 91, 51–58.

Yamashita, Y., Maki, A., Koizumi, H., 2001. Wavelength dependence of the precisionof noninvasive optical measurement of oxy-, deoxy-, and total-hemoglobinconcentration. Med. Phys. 28, 1108–1114.

Yerkes, R.M., Dodson, J.D., 1908. The relation of strength of stimulus to rapidity ofhabit-formation. J. Comp. Neurol. Psychol. 18, 459–482.

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566

676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132



neuropsychologia - university of victoriaweb.uvic.ca/.../exhaustingexercise.pdf · 2015-01-07 ·...

Documents